Temperature sensor: from control to optimisation
Temperature is a crucial parameter in industry, research, security and environment since an excess of temperature is usually associated with ageing,erratic behaviours, and then failure. Therefore, temperature has to be monitored, or systems have to be designed as to avoid uncontrolled heating.
Most industrial systems include one or more temperature sensors. These sensors monitor the temperature in real-time, so that when above or below a threshold a regulation system is used to, respectively, reduce or increase temperature. In some critical situation, an emergency system can be activated. An example can be found in boilers but also in nuclear thermal power plant. Different sensor technologies have been developed for this purpose, for instance based on semiconductor junctions, thermo-mechanical effects, resistivity variations, Seebeck, Thomson, or Peltier effects. These technologies have the advantages to be cheap, robust, easy to use and reliable, but need a contact to make a measurement. Black body radiations, in the infrared for temperatures below 3,600K according to Wien, can be used for contactless measurements and also thermal imaging. Although much more expensive, infrared technology allows industrial processes to be optimized such as in glass industry where a high quality glass is obtained by a controlled cooling. Also, intrusions can be detected in strategic sites or habitation thermal losses can be estimated. A less expected utilization of infrared imaging systems was the control of passenger temperature in airports during the avian flu pandemic.
The development of new materials and optimisation of devices intrinsically rely on the knowledge and the management of heat flux behaviour at micro and even nanoscale. It has been indeed reported (Domingues et al., 2005) that inclusion of nanoparticles in a material increases drastically its thermal properties without modifying its electrical properties. It is also known that integrated circuits could work at higher operating frequency and drive more power with a better thermal management. In both cases, high spatial resolution thermal imaging systems and accurate temperature measurements are needed in order to validate theoretical dissipation models and integrated circuit designs.
Different contactless approaches are being developed in order to fulfil these requirements. As a first example, a very good accuracy can be obtained with a mirage effect setup (Paoloni and Bertolotti, 2003). In this setup, a laser pump beam heats the surface of the material to be tested which produces a temperature gradient in the air above the material. A laser probe beam propagating in this region is then deflected depending on the material thermal diffusivity. As a second example, high spatial resolution can be obtained with photoreflectance setup (Tessier et al., 2005). In this setup, the sample under test is illuminated by a monochromatic light. Since, reflectivity depends on temperature, the difference between an image of an integrated circuit under operation and an image of the same integrated circuit at rest directly provides the thermal map of that circuit. The advantage is that spatial resolution is only limited by the illumination wavelength and can be as small as 350nm with visible light.
The spatial resolution can be further improved using near- field sensing (Guo et al., 2006). In this case, a thermal microprobe which can be a microthermocouple or a micro- resistor, is displaced on the material surface. The spatial resolution depends nevertheless on the roughness of the surface.
In conclusion, each setup has its own advantages and is well suited for a particular application. With the continuous trend of system miniaturization, new thermal sensors should be developed in order to still improve their spatial resolution, signal to noise ratio and acquisition time.
Céline FilloyUniversité Pierre et Marie Curie, Paris, France, CNRS, Paris, France and ESPCI, Paris, France
